Stem Separation
Isolate vocals, drums, bass, and other instruments from any audio track with AI-powered precision
Overview
Stem separation (also called source separation) uses advanced AI to split mixed audio into individual components: vocals, drums, bass, and other instruments. This powerful feature enables remixing, sampling, analysis, and creative manipulation of any audio track.
How to Access Stem Separation
Stem separation is available directly in the Wubble interface. On the chat page, you'll find a "Stems" button on each audio card. Click it to separate your track into individual stems. The process typically takes 30-90 seconds.
Not Available via API
Stem separation is currently only available through the Wubble web interface and is not accessible via API endpoints.
Vocals
Lead vocals, backing vocals, harmonies
Drums
Kick, snare, hi-hats, cymbals, toms
Bass
Bass guitar, synth bass, sub bass
Other
Keys, guitars, strings, brass, effects
What You Can Do
How It Works
Wubble uses state-of-the-art deep learning models trained on millions of songs to understand and separate audio sources. The AI has learned to identify the unique characteristics of different instruments and vocals, even when they overlap in frequency and time.
Audio Analysis
The AI analyzes your audio in the time-frequency domain, examining spectral characteristics, temporal patterns, and harmonic structures
Source Identification
Deep neural networks identify which parts of the mixed audio correspond to vocals, drums, bass, and other instruments
Intelligent Separation
The AI creates separate audio streams for each source, using sophisticated algorithms to minimize artifacts and bleed
Quality Enhancement
Post-processing removes artifacts, reduces noise, and ensures phase coherence for professional-quality results
Cutting-Edge Technology
Wubble uses the latest advances in AI music understanding, achieving studio-quality separation that rivals professional manual isolation.
Available Stems
Wubble separates your audio into 4 standard stems, covering the most common categories used in music production. This is perfect for remixing, sampling, and creative applications.
Standard Stems
Vocals
All vocal content including lead vocals, backing vocals, harmonies, ad-libs, and vocal effects. Perfect for creating acapellas or removing vocals entirely.
Drums
All percussive elements: kick drums, snares, hi-hats, cymbals, toms, percussion, and drum fills. Ideal for isolating rhythm sections or creating new drum mixes.
Bass
Low-frequency content including bass guitar, synth bass, sub bass, and bass synths. Essential for isolating the bottom end of a mix.
Other
Everything else: guitars, keyboards, strings, brass, woodwinds, synths, sound effects, and ambient elements. Contains all harmonic and melodic non-drum content.
How to Separate Stems
Follow these simple steps to separate your audio into individual stems:
Navigate to Chat Page
Go to your chat page where your generated audio tracks are displayed as audio cards
Click the Stems Button
On each audio card, you'll find a "Stems" button. Click it to initiate stem separation for that track
Wait for Processing
The AI will process your track, typically taking 30-90 seconds depending on track length and complexity
Download Your Stems
Once complete, you can download individual stems (vocals, drums, bass, other) for use in your projects
Processing Time
Most tracks complete stem separation in 30-90 seconds. Longer or more complex tracks may take slightly more time. You'll see a progress indicator while processing.
Use Cases & Applications
🎧 Music Production & Remixing
Create official remixes, mashups, and bootlegs. Isolate stems to build entirely new arrangements or add your own elements to existing tracks.
Example: Extract vocals from popular tracks to create DJ mashups and remixes
🎓 Music Education
Isolate specific instruments for transcription, learning, and analysis. Study production techniques by examining individual stems.
Example: Isolate bass lines from jazz recordings to learn complex walking bass patterns
🎤 Karaoke & Practice
Create professional karaoke tracks by removing vocals. Musicians can practice along with isolated instrument tracks.
Example: Remove vocals from any song to create custom karaoke versions
🎬 Content Creation
Remove vocals from songs for YouTube videos, podcasts, and other content to avoid copyright issues while keeping the music.
Example: Use instrumental versions of popular songs as background music
🎵 Sampling & Beat Making
Extract clean drum loops, bass lines, or melodic elements for use in your own productions. Create sample packs from existing music.
Example: Isolate vintage drum breaks for hip-hop production
🎚️ Audio Restoration & Forensics
Isolate and enhance specific elements in archival recordings or analyze audio for forensic purposes.
Example: Extract and restore vocals from old concert recordings
Best Practices
Start with Generated Tracks
Stem separation works best on tracks generated within Wubble, as our AI has full context of the musical structure and elements.
Allow Time for Processing
Stem separation typically takes 30-90 seconds. Don't navigate away from the page while processing is in progress.
Download Individual Stems
Once separation is complete, you can download each stem individually. Choose the stems you need for your specific project.
Understand Natural Limitations
Some bleed between stems is normal, especially in densely mixed tracks. This is a characteristic of AI-based separation technology.
Use for Creative Projects
Stem separation is perfect for remixing, sampling, learning, and creative manipulation of your Wubble-generated audio.
Check Phase When Recombining
When recombining stems in your DAW, ensure they're phase-aligned. Wubble maintains phase relationships so stems sum back to the original when combined.